Scatter plot regions to image space and viceversa

Is there a way in SNAP to define regions of interest in the scatter plot and reflect those regions in the image space? Other software has that capability and it is very useful. To move from scatter plot space to the image space, for instance highlighting all the pixels in the image that are within in the region of the scatter plot. Or the other way around, to identify pixels in the image and they would be highlited in the scatter plot.



That’s possible. Do a right-click on the scatterplot and chose Select Mask ‘scatter_plot_area’ from the menu.


I didn’t know of this feature yet. Very useful indeed. And very valuable for teaching remote sensing.

I just tested it - great! I love it.


Most of the scatter plots look like a tasseled cap transform.

Thanks for your prompt response, I am glad snap has this feature, scatter to image, but the other way around from image to scatter?.

Could you inform me what does the dl units mean? exploring the jp2 S2 images in the SAFE format they have unsigned int 16 bit cells, but when displayed are dipslapey (histogram and scattter plot) they are presented as floating numbers (reflectance?) and have dl in the units. When did this change happened, through the metadata?

In Abraun tests-examples, the bands are presented as radiance in W m^2 sr um and in others as nm. How can you control the displayed units in SNAP.


the scatter plot displays the units stored in the metadata. I used uncalibrated Landsat TM data. dl in marpets data means dimensionless. Sentinel-2 L1C products show Top Of Atmosphere radiance.
If you calibrate your data to L2A, the plot shows reflectance (most suitable for scatter plots).

Even this is possible. Draw a rectangle or some polygone and use it as ROI in the scatterplot.

The units are taken from the bands. Usually they are specified by the data. But it can be changed in the property dialog of the band. Right-click on a band select Properties from the menu and then change the unit.

1 Like